data.py 18.8 KB
Newer Older
Markus Scheidgen's avatar
Markus Scheidgen committed
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
# Copyright 2018 Markus Scheidgen
#
# Licensed under the Apache License, Version 2.0 (the "License");
# you may not use this file except in compliance with the License.
# You may obtain a copy of the License at
#
#   http://www.apache.org/licenses/LICENSE-2.0
#
# Unless required by applicable law or agreed to in writing, software
# distributed under the License is distributed on an"AS IS" BASIS,
# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
# See the License for the specific language governing permissions and
# limitations under the License.

"""
This module comprises a set of persistent document classes that hold all user related
data. These are information about users, their uploads and datasets, the associated
calculations, and files


.. autoclass:: Calc
    :members:
.. autoclass:: Upload
    :members:
"""

from typing import List, Any
from datetime import datetime
29
from elasticsearch.exceptions import NotFoundError
Markus Scheidgen's avatar
Markus Scheidgen committed
30
from mongoengine import StringField, BooleanField, DateTimeField, DictField, IntField
Markus Scheidgen's avatar
Markus Scheidgen committed
31
import logging
32
import base64
Markus Scheidgen's avatar
Markus Scheidgen committed
33
import time
34
from structlog import wrap_logger
Markus Scheidgen's avatar
Markus Scheidgen committed
35

36
from nomad import config, utils
Markus Scheidgen's avatar
Markus Scheidgen committed
37
from nomad.files import UploadFile, ArchiveFile, ArchiveLogFile, File
Markus Scheidgen's avatar
Markus Scheidgen committed
38
from nomad.repo import RepoCalc
Markus Scheidgen's avatar
Markus Scheidgen committed
39
from nomad.user import User
40
from nomad.processing.base import Proc, Chord, process, task, PENDING, SUCCESS, FAILURE, RUNNING
Markus Scheidgen's avatar
Markus Scheidgen committed
41
from nomad.parsing import parsers, parser_dict
Markus Scheidgen's avatar
Markus Scheidgen committed
42
from nomad.normalizing import normalizers
Markus Scheidgen's avatar
Markus Scheidgen committed
43
from nomad.utils import lnr
Markus Scheidgen's avatar
Markus Scheidgen committed
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73


class NotAllowedDuringProcessing(Exception): pass


class Calc(Proc):
    """
    Instances of this class represent calculations. This class manages the elastic
    search index entry, files, and archive for the respective calculation.

    It also contains the calculations processing and its state.

    The attribute list, does not include the various repository properties generated
    while parsing, including ``program_name``, ``program_version``, etc.

    Attributes:
        archive_id: the hash based archive id of the calc
        parser: the name of the parser used to process this calc
        upload_id: the id of the upload used to create this calculation
        mainfile: the mainfile (including path in upload) that was used to create this calc
        mainfile_tmp_path: path to the mainfile extracted for processing
    """
    archive_id = StringField(primary_key=True)
    upload_id = StringField()
    mainfile = StringField()
    parser = StringField()
    mainfile_tmp_path = StringField()

    meta: Any = {
        'indices': [
74
            'upload_id', 'mainfile', 'code', 'parser', 'status'
Markus Scheidgen's avatar
Markus Scheidgen committed
75
76
77
78
79
80
81
        ]
    }

    def __init__(self, *args, **kwargs):
        super().__init__(*args, **kwargs)
        self._parser_backend = None
        self._upload = None
82
        self._calc_proc_logwriter = None
Markus Scheidgen's avatar
Markus Scheidgen committed
83
        self._calc_proc_logfile = None
Markus Scheidgen's avatar
Markus Scheidgen committed
84
85
86
87
88

    @classmethod
    def get(cls, id):
        return cls.get_by_id(id, 'archive_id')

Markus Scheidgen's avatar
Markus Scheidgen committed
89
90
91
92
    @property
    def mainfile_file(self) -> File:
        return File(self.mainfile_tmp_path)

Markus Scheidgen's avatar
Markus Scheidgen committed
93
94
95
96
    def delete(self):
        """
        Delete this calculation and all associated data. This includes all files,
        the archive, and this search index entry.
97
        TODO is this needed? Or do we always delete hole uploads in bulk.
Markus Scheidgen's avatar
Markus Scheidgen committed
98
99
100
        """
        # delete the archive
        if self.archive_id is not None:
101
            ArchiveFile(self.archive_id).delete()
Markus Scheidgen's avatar
Markus Scheidgen committed
102
103

        # delete the search index entry
104
105
106
107
108
109
        try:
            elastic_entry = RepoCalc.get(self.archive_id)
            if elastic_entry is not None:
                elastic_entry.delete()
        except NotFoundError:
            pass
Markus Scheidgen's avatar
Markus Scheidgen committed
110
111
112
113
114
115
116
117
118

        # delete this mongo document
        super().delete()

    def get_logger(self, **kwargs):
        upload_hash, calc_hash = self.archive_id.split('/')
        logger = super().get_logger()
        logger = logger.bind(
            upload_id=self.upload_id, mainfile=self.mainfile,
119
            upload_hash=upload_hash, calc_hash=calc_hash,
Markus Scheidgen's avatar
Markus Scheidgen committed
120
            archive_id=self.archive_id, **kwargs)
121

Markus Scheidgen's avatar
Markus Scheidgen committed
122
123
        return logger

124
125
126
127
128
129
130
131
    def get_calc_logger(self, **kwargs):
        """
        Returns a wrapped logger that additionally saves all entries to the calculation
        processing log in the archive.
        """
        logger = self.get_logger(**kwargs)

        if self._calc_proc_logwriter is None:
Markus Scheidgen's avatar
Markus Scheidgen committed
132
133
            self._calc_proc_logfile = ArchiveLogFile(self.archive_id)
            self._calc_proc_logwriter = self._calc_proc_logfile.open('wt')
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148

        def save_to_cacl_log(logger, method_name, event_dict):
            program = event_dict.get('normalizer', 'parser')
            event = event_dict.get('event', '')
            entry = '[%s] %s: %s' % (method_name, program, event)
            if len(entry) > 120:
                self._calc_proc_logwriter.write(entry[:120])
                self._calc_proc_logwriter.write('...')
            else:
                self._calc_proc_logwriter.write(entry)
            self._calc_proc_logwriter.write('\n')
            return event_dict

        return wrap_logger(logger, processors=[save_to_cacl_log])

Markus Scheidgen's avatar
Markus Scheidgen committed
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
    @property
    def json_dict(self):
        """ A json serializable dictionary representation. """
        data = {
            'archive_id': self.archive_id,
            'mainfile': self.mainfile,
            'upload_id': self.upload_id,
            'parser': self.parser
        }
        data.update(super().json_dict)
        return {key: value for key, value in data.items() if value is not None}

    @process
    def process(self):
        self._upload = Upload.get(self.upload_id)
164
        logger = self.get_logger()
Markus Scheidgen's avatar
Markus Scheidgen committed
165
        if self._upload is None:
166
            logger.error('calculation upload does not exist')
Markus Scheidgen's avatar
Markus Scheidgen committed
167
168
169
170
171
172

        try:
            self.parsing()
            self.normalizing()
            self.archiving()
        finally:
Markus Scheidgen's avatar
Markus Scheidgen committed
173
            # close loghandler that was not closed due to failures
174
            try:
175
176
177
                if self._calc_proc_logwriter is not None:
                    self._calc_proc_logwriter.close()
                    self._calc_proc_logwriter = None
178
179
180
181
            except Exception as e:
                logger.error('could not close calculation proc log', exc_info=e)

            # inform parent proc about completion
182
            self._upload.completed_child()
Markus Scheidgen's avatar
Markus Scheidgen committed
183
184
185

    @task
    def parsing(self):
186
187
        context = dict(parser=self.parser, step=self.parser)
        logger = self.get_calc_logger(**context)
188
        parser = parser_dict[self.parser]
Markus Scheidgen's avatar
Markus Scheidgen committed
189

190
        with utils.timer(logger, 'parser executed', input_size=self.mainfile_file.size):
191
            self._parser_backend = parser.run(self.mainfile_tmp_path, logger=logger)
Markus Scheidgen's avatar
Markus Scheidgen committed
192

Markus Scheidgen's avatar
Markus Scheidgen committed
193
        if self._parser_backend.status[0] != 'ParseSuccess':
194
            logger.error(self._parser_backend.status[1])
Markus Scheidgen's avatar
Markus Scheidgen committed
195
            error = self._parser_backend.status[1]
196
            self.fail(error, level=logging.DEBUG, **context)
Markus Scheidgen's avatar
Markus Scheidgen committed
197
198
199
200
201

    @task
    def normalizing(self):
        for normalizer in normalizers:
            normalizer_name = normalizer.__name__
202
203
            context = dict(normalizer=normalizer_name, step=normalizer_name)
            logger = self.get_calc_logger(**context)
Markus Scheidgen's avatar
Markus Scheidgen committed
204
205

            with utils.timer(
206
                    logger, 'normalizer executed', input_size=self.mainfile_file.size):
207
                normalizer(self._parser_backend).normalize(logger=logger)
Markus Scheidgen's avatar
Markus Scheidgen committed
208

Markus Scheidgen's avatar
Markus Scheidgen committed
209
            if self._parser_backend.status[0] != 'ParseSuccess':
210
                logger.error(self._parser_backend.status[1])
Markus Scheidgen's avatar
Markus Scheidgen committed
211
                error = self._parser_backend.status[1]
212
                self.fail(error, level=logging.WARNING, **context)
Markus Scheidgen's avatar
Markus Scheidgen committed
213
                return
214
            logger.debug(
Markus Scheidgen's avatar
Markus Scheidgen committed
215
216
217
218
                'completed normalizer successfully', normalizer=normalizer_name)

    @task
    def archiving(self):
219
220
221
222
223
224
225
226
227
228
229
        logger = self.get_logger()

        upload_hash, calc_hash = self.archive_id.split('/')
        additional = dict(
            mainfile=self.mainfile,
            upload_time=self._upload.upload_time,
            staging=True,
            restricted=False,
            user_id=self._upload.user_id)

        with utils.timer(logger, 'indexed', step='index'):
230
231
232
233
234
235
236
237
            # persist to elastic search
            RepoCalc.create_from_backend(
                self._parser_backend,
                additional=additional,
                upload_hash=upload_hash,
                calc_hash=calc_hash,
                upload_id=self.upload_id)

Markus Scheidgen's avatar
Markus Scheidgen committed
238
239
240
241
        with utils.timer(
                logger, 'archived', step='archive',
                input_size=self.mainfile_file.size) as log_data:

242
            # persist the archive
Markus Scheidgen's avatar
Markus Scheidgen committed
243
244
            archive_file = ArchiveFile(self.archive_id)
            with archive_file.write_archive_json() as out:
245
246
                self._parser_backend.write_json(out, pretty=True)

Markus Scheidgen's avatar
Markus Scheidgen committed
247
248
249
250
251
252
253
            log_data.update(archive_size=archive_file.size)

        # close loghandler
        if self._calc_proc_logwriter is not None:
            with utils.timer(
                    logger, 'archived log', step='archive_log',
                    input_size=self.mainfile_file.size) as log_data:
254
255
                self._calc_proc_logwriter.close()
                self._calc_proc_logwriter = None
Markus Scheidgen's avatar
Markus Scheidgen committed
256

Markus Scheidgen's avatar
Markus Scheidgen committed
257
258
                log_data.update(log_size=self._calc_proc_logfile.size)

Markus Scheidgen's avatar
Markus Scheidgen committed
259

260
class Upload(Chord):
Markus Scheidgen's avatar
Markus Scheidgen committed
261
262
263
264
265
266
    """
    Represents uploads in the databases. Provides persistence access to the files storage,
    and processing state.

    Attributes:
        name: optional user provided upload name
267
        local_path: optional local path, e.g. for files that are already somewhere on the server
Markus Scheidgen's avatar
Markus Scheidgen committed
268
269
270
271
272
273
        additional_metadata: optional user provided additional meta data
        upload_id: the upload id generated by the database
        in_staging: true if the upload is still in staging and can be edited by the uploader
        is_private: true if the upload and its derivitaves are only visible to the uploader
        upload_time: the timestamp when the system realised the upload
        upload_hash: the hash of the uploaded file
274
        user_id: the id of the user that created this upload
Markus Scheidgen's avatar
Markus Scheidgen committed
275
276
277
278
279
280
    """
    id_field = 'upload_id'

    upload_id = StringField(primary_key=True)

    name = StringField(default=None)
281
    local_path = StringField(default=None)
Markus Scheidgen's avatar
Markus Scheidgen committed
282
283
284
285
286
287
288
289
    additional_metadata = DictField(default=None)

    in_staging = BooleanField(default=True)
    is_private = BooleanField(default=False)

    upload_time = DateTimeField()
    upload_hash = StringField(default=None)

290
    user_id = StringField(required=True)
291
292
    upload_url = StringField(default=None)
    upload_command = StringField(default=None)
Markus Scheidgen's avatar
Markus Scheidgen committed
293

294
295
    _initiated_parsers = IntField(default=-1)

Markus Scheidgen's avatar
Markus Scheidgen committed
296
297
    meta: Any = {
        'indexes': [
298
            'upload_hash', 'user_id', 'status'
Markus Scheidgen's avatar
Markus Scheidgen committed
299
300
301
302
303
304
305
306
307
308
309
310
311
312
        ]
    }

    def __init__(self, **kwargs):
        super().__init__(**kwargs)
        self._upload = None

    @classmethod
    def get(cls, id):
        return cls.get_by_id(id, 'upload_id')

    @classmethod
    def user_uploads(cls, user: User) -> List['Upload']:
        """ Returns all uploads for the given user. Currently returns all uploads. """
Markus Scheidgen's avatar
Markus Scheidgen committed
313
        return cls.objects(user_id=user.email, in_staging=True)
Markus Scheidgen's avatar
Markus Scheidgen committed
314
315
316
317
318
319
320
321
322
323
324
325
326
327

    def get_logger(self, **kwargs):
        logger = super().get_logger()
        logger = logger.bind(upload_id=self.upload_id, **kwargs)
        return logger

    def delete(self):
        logger = self.get_logger(task='delete')

        if not (self.completed or self.is_stale or self.current_task == 'uploading'):
            raise NotAllowedDuringProcessing()

        with lnr(logger, 'delete upload file'):
            try:
328
                UploadFile(self.upload_id, local_path=self.local_path).delete()
Markus Scheidgen's avatar
Markus Scheidgen committed
329
330
331
332
333
334
335
336
337
            except KeyError:
                if self.current_task == 'uploading':
                    logger.debug(
                        'Upload exist, but file does not exist. '
                        'It was probably aborted and deleted.')
                else:
                    logger.debug('Upload exist, but uploaded file does not exist.')

        with lnr(logger, 'deleting calcs'):
338
            # delete archive files
339
            ArchiveFile.delete_archives(upload_hash=self.upload_hash)
340
341

            # delete repo entries
Markus Scheidgen's avatar
Markus Scheidgen committed
342
            RepoCalc.delete_upload(upload_id=self.upload_id)
343
344
345

            # delete calc processings
            Calc.objects(upload_id=self.upload_id).delete()
Markus Scheidgen's avatar
Markus Scheidgen committed
346
347
348
349

        with lnr(logger, 'deleting upload'):
            super().delete()

350
351
    @classmethod
    def _external_objects_url(cls, url):
352
        """ Replaces the given internal object storage url with an URL that allows
353
354
            external access.
        """
355
        return 'http://%s:%s%s%s' % (config.services.api_host, config.services.api_port, config.services.api_base_path, url)
356

Markus Scheidgen's avatar
Markus Scheidgen committed
357
358
359
360
361
362
    @classmethod
    def create(cls, **kwargs) -> 'Upload':
        """
        Creates a new upload for the given user, a user given name is optional.
        It will populate the record with a signed url and pending :class:`UploadProc`.
        The upload will be already saved to the database.
363
364
365

        Arguments:
            user (User): The user that created the upload.
Markus Scheidgen's avatar
Markus Scheidgen committed
366
        """
367
368
369
370
371
        user: User = kwargs['user']
        del(kwargs['user'])
        if 'upload_id' not in kwargs:
            kwargs.update(upload_id=utils.create_uuid())
        kwargs.update(user_id=user.email)
Markus Scheidgen's avatar
Markus Scheidgen committed
372
        self = super().create(**kwargs)
373
374
375
376

        basic_auth_token = base64.b64encode(b'%s:' % user.generate_auth_token()).decode('utf-8')

        self.upload_url = cls._external_objects_url('/uploads/%s/file' % self.upload_id)
Markus Scheidgen's avatar
Markus Scheidgen committed
377
        self.upload_command = 'curl -H "Authorization: Basic %s" "%s" --upload-file local_file' % (
378
379
            basic_auth_token, self.upload_url)

Markus Scheidgen's avatar
Markus Scheidgen committed
380
        self._continue_with('uploading')
381

Markus Scheidgen's avatar
Markus Scheidgen committed
382
383
384
385
386
387
388
389
390
        return self

    @property
    def is_stale(self) -> bool:
        if self.current_task == 'uploading' and self.upload_time is None:
            return (datetime.now() - self.create_time).days > 1
        else:
            return False

Markus Scheidgen's avatar
Markus Scheidgen committed
391
    def unstage(self):
392
        self.get_logger().info('unstage')
Markus Scheidgen's avatar
Markus Scheidgen committed
393
        self.in_staging = False
394
        RepoCalc.unstage(upload_id=self.upload_id)
Markus Scheidgen's avatar
Markus Scheidgen committed
395
396
        self.save()

Markus Scheidgen's avatar
Markus Scheidgen committed
397
398
399
400
401
    @property
    def json_dict(self) -> dict:
        """ A json serializable dictionary representation. """
        data = {
            'name': self.name,
402
            'local_path': self.local_path,
Markus Scheidgen's avatar
Markus Scheidgen committed
403
404
            'additional_metadata': self.additional_metadata,
            'upload_id': self.upload_id,
405
            'upload_url': self.upload_url,
406
            'upload_command': self.upload_command,
Markus Scheidgen's avatar
Markus Scheidgen committed
407
408
409
410
411
412
413
414
415
416
417
418
419
420
421
422
423
424
425
            'upload_time': self.upload_time.isoformat() if self.upload_time is not None else None,
            'is_stale': self.is_stale,
        }
        data.update(super().json_dict)
        return {key: value for key, value in data.items() if value is not None}

    @process
    def process(self):
        self.extracting()
        self.parse_all()

    @task
    def uploading(self):
        pass

    @task
    def extracting(self):
        logger = self.get_logger()
        try:
Markus Scheidgen's avatar
Markus Scheidgen committed
426
427
428
429
            self._upload = UploadFile(self.upload_id, local_path=self.local_path)
            with utils.timer(
                    logger, 'upload extracted', step='extracting',
                    upload_size=self._upload.size):
430
                self._upload.extract()
Markus Scheidgen's avatar
Markus Scheidgen committed
431
432
433
434
435
436
        except KeyError as e:
            self.fail('process request for non existing upload', level=logging.INFO)
            return

        try:
            self.upload_hash = self._upload.hash()
437
        except Exception as e:
Markus Scheidgen's avatar
Markus Scheidgen committed
438
439
440
            self.fail('could not create upload hash', e)
            return

Markus Scheidgen's avatar
Markus Scheidgen committed
441
        if RepoCalc.upload_exists(self.upload_hash):
Markus Scheidgen's avatar
Markus Scheidgen committed
442
443
444
445
446
            self.fail('The same file was already uploaded and processed.', level=logging.INFO)
            return

    @task
    def parse_all(self):
447
448
        logger = self.get_logger()

Markus Scheidgen's avatar
Markus Scheidgen committed
449
        # TODO: deal with multiple possible parser specs
Markus Scheidgen's avatar
Markus Scheidgen committed
450
451
452
453
        with utils.timer(
                logger, 'upload extracted', step='matching',
                upload_size=self._upload.size,
                upload_filecount=len(self._upload.filelist)):
454
455
456
457
            total_calcs = 0
            for filename in self._upload.filelist:
                for parser in parsers:
                    try:
Markus Scheidgen's avatar
Markus Scheidgen committed
458
459
460
                        potential_mainfile = self._upload.get_file(filename)
                        if parser.is_mainfile(filename, lambda fn: potential_mainfile.open()):
                            mainfile_path = potential_mainfile.os_path
461
462
463
                            calc = Calc.create(
                                archive_id='%s/%s' % (self.upload_hash, utils.hash(filename)),
                                mainfile=filename, parser=parser.name,
Markus Scheidgen's avatar
Markus Scheidgen committed
464
                                mainfile_tmp_path=mainfile_path,
465
466
467
468
469
                                upload_id=self.upload_id)

                            calc.process()
                            total_calcs += 1
                    except Exception as e:
Markus Scheidgen's avatar
Markus Scheidgen committed
470
                        self.error(
471
472
                            'exception while matching pot. mainfile',
                            mainfile=filename, exc_info=e)
Markus Scheidgen's avatar
Markus Scheidgen committed
473

474
475
476
477
478
        # have to save the total_calcs information for chord management
        self.spwaned_childred(total_calcs)

    def join(self):
        self.cleanup()
Markus Scheidgen's avatar
Markus Scheidgen committed
479
480
481
482

    @task
    def cleanup(self):
        try:
Markus Scheidgen's avatar
Markus Scheidgen committed
483
484
485
486
487
            upload = UploadFile(self.upload_id, local_path=self.local_path)
            with utils.timer(
                    self.get_logger(), 'processing cleaned up', step='cleaning',
                    upload_size=upload.size):
                upload.remove_extract()
Markus Scheidgen's avatar
Markus Scheidgen committed
488
        except KeyError as e:
Markus Scheidgen's avatar
Markus Scheidgen committed
489
            self.fail('Upload does not exist', exc_info=e)
Markus Scheidgen's avatar
Markus Scheidgen committed
490
491
492
493
494
            return

        self.get_logger().debug('closed upload')

    @property
495
496
497
498
499
500
501
502
503
504
505
    def processed_calcs(self):
        return Calc.objects(upload_id=self.upload_id, status__in=[SUCCESS, FAILURE]).count()

    @property
    def total_calcs(self):
        return Calc.objects(upload_id=self.upload_id).count()

    @property
    def failed_calcs(self):
        return Calc.objects(upload_id=self.upload_id, status=FAILURE).count()

506
507
508
509
    @property
    def pending_calcs(self):
        return Calc.objects(upload_id=self.upload_id, status=PENDING).count()

510
511
    def all_calcs(self, start, end, order_by='mainfile'):
        return Calc.objects(upload_id=self.upload_id)[start:end].order_by(order_by)
512

Markus Scheidgen's avatar
Markus Scheidgen committed
513
    @staticmethod
514
515
516
517
518
    def repair_all():
        """
        Utitlity function that will look for suspiciously looking conditions in
        all uncompleted downloads. It ain't a perfect world.
        """
519
520
521
        # TODO this was added as a quick fix to #37.
        # Even though it might be strictly necessary, there should be a tested backup
        # solution for it Chords to not work properly due to failed to fail processings
522
523
524
525
526
527
528
529
530
531
        uploads = Upload.objects(status__in=[PENDING, RUNNING])
        for upload in uploads:
            completed = upload.processed_calcs
            total = upload.total
            pending = upload.pending_calcs

            if completed + pending == total:
                time.sleep(2)
                if pending == upload.pending_calcs:
                    Calc.objects(upload_id=upload.upload_id, status=PENDING).delete()